Mendelian Randomization

Mendelian Randomization in R

Overview

Mendelian randomization (MR) methods for causal inference using genetic variants as instrumental variables. Covers instrument selection, two-sample MR, sensitivity analyses, pleiotropy assessment, multivariable MR, and advanced methods for robust causal inference.

Instrument Selection

Using TwoSampleMR

library(TwoSampleMR)

# Extract instruments from GWAS database
# IEU Open GWAS Project
exposure_dat 
  filter(pval.exposure  10 indicates strong instruments

calculate_f_stat 
  filter(abs(b - results$b[results$method == "Inverse variance weighted"]) >
         2 * results$se[results$method == "Inverse variance weighted"])

MR-PRESSO

library(MRPRESSO)

# MR-PRESSO for outlier detection
presso 
  filter(pval_exp 
  pull(snp)

# Fit CAUSE
cause_result  Y
forward_results  X
# Extract instruments for outcome
outcome_instruments  Y", "Y -> X"),
  Beta = c(forward_results$b[1], reverse_results$b[1]),
  SE = c(forward_results$se[1], reverse_results$se[1]),
  P = c(forward_results$pval[1], reverse_results$pval[1])
)
print(comparison)

Reporting Results

# Create comprehensive MR report
create_mr_report 
      select(method, nsnp, b, se, pval) |>
      mutate(
        or = exp(b),
        ci_lower = exp(b - 1.96 * se),
        ci_upper = exp(b + 1.96 * se)
      ),
    heterogeneity = het,
    pleiotropy = pleiotropy,
    f_statistic = mean(dat$f_stat, na.rm = TRUE)
  )

  return(report)
}

# Generate report
mr_report  10 for all instruments
2. **Multiple methods**: Report IVW, MR-Egger, weighted median, and MR-PRESSO
3. **Sensitivity analyses**: Always check heterogeneity, pleiotropy, and leave-one-out
4. **Steiger filtering**: Assess support for the assumed direction; it does not prove direction
5. **Visualization**: Include scatter, forest, and funnel plots
6. **Bidirectional**: Consider reverse causation when biologically plausible
7. **Biological plausibility**: Interpret results in biological context
8. **Reporting**: Follow STROBE-MR guidelines

## Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

- **Author:** [choxos](https://github.com/choxos)
- **Source:** [choxos/BiostatAgent](https://github.com/choxos/BiostatAgent)
- **License:** MIT

Install and usage instructions live in the source repository linked above.